intech-free space optical technologies

7/28/2019 InTech-Free Space Optical Technologies

1/41

Free Space Optical Technologies 257

X

Free Space Optical Technologies

Davide M. Forin, G. IncertiUniversity of Rome Tor Vergata Italy

G.M. Tosi BeleffiISCOM, Italian Ministry of Economic Development Comm. Department, Rome Italy

A.L.J. Teixeira, L.N. Costa, P.S. De Brito AndrInstituto de Telecomunicaes, Aveiro, Portugal

B. Geiger, E. Leitgeb and F. NadeemGraz University of Technology

1. Introduction (D. Forin G.M. Tosi Beleffi B. Geiger E. Leitgeb)

Free Space Optics (FSO), also known as Optical Wireless or Lasercom (i.e. LaserCommunications), is a re-emerging technology using modulated optical beams to establishshort, medium or long reach wireless data transmission. Most of the attention on FSOcommunication systems it was initially boost by military purposes and first development ofthis technology was dedicated to the solution of issues related to defense applications.

Todays market interest to FSO refers to both civil and military scenarios covering differentsituations and different environments, from undersea to space. In particular, due to the highcarrier frequency of 300 THz and the consequently high bandwidth, the most prominentadvantage of Free Space Optical (FSO) communication links may be their potential for veryhigh data rates of several Gbps (up to 40 Gbps in the future (J. Wells, 2009)). Otheradvantages like license-free operation, easy installation, commercial availability, andinsensitivity to electromagnetic interference, jamming, or wiretapping make FSO interestingfor applications like last mile access, airborne and satellite communication (L. Stotts et alt,2009), temporary mobile links and permanent connections between buildings. Mainly, theadoption of FSO is needed when a physical connection is not a practicable solution andwhere is requested to handle an high bandwidth. As a matter of fact, FSO is the only

technology, in the wireless scenario, able to grant bandwidth of several Gigabits per second.The interest in this technology is also due to the low initial CAPEX (Capital Expenditure)requirement, to the intrinsic high-level data protection & security, to the good flexibility andgreat scalability innate in this solution. For these reasons FSO possible applications covertoday, as mentioned, a wide range. Thus this technology generates interest in severalmarkets: the first/last mile in dense urban areas, network access for isolated premises, high-speed LAN-to-LAN (Local Area Networks) and even chip-to-chip connections, transitionaland temporary network connection, undersea and space communication. Furthermore FSOcan be used as an alternative or upgrade add-on to existing wireless technologies when theclimatic conditions permit its full usage.

13

www.intechopen.com


2/41

Trends in Telecommunications Technologies258

In spite of the growing interest in space and undersea applications, infact, the terrestrial FSOstill remains of primary significance and the performance of such links is highly dependenton different weather conditions. Atmospheric effects affect the distance and the availabilityof the optical wireless links so not all the geographical sites are suitable for this kind of

broadband solution. Links as long as 7 km are in operation, but prior to the deployment ofwireless optical links the average weather condition must be evaluated to estimate theexpected outage time on the link in that area. The outage will depend on the link length andon the persistence of adverse weather conditions. So in general, it can be affirmed that shortenough links, hundreds of meters, will be operational also with the worst possible weatherconditions.Besides commercially available opaque systems, where the optical signal is terminated to anelectronic receiver and subsequently sent through the atmosphere by means of a dedicatedlaser, a new configuration, known as fully transparent, is under study. The bandwidthachievable in these last systems is comparable to the optical fiber one. Being absent any kindof optical to electrical bottleneck (OEO). For transparency, infact, is meant launching andcollecting power directly through single mode optical fibers. Such new extremely highbandwidth wireless systems, although still in the research stage, are gaining more and moreinterest especially in the last mile scenarios. Last but not least, compared to a microwavelink, an FSO link can support higher bit rates and its operational frequencies are license-freein all jurisdictions apart for an year low cost fee that must be paid to the reference PA(Public Administration). The Authors want to outline that the work carried out in thischapter has been done in the framework of the European funded FP7 NoE BONE Project(WP13) and the COST IC0802 Action.

2. History of Free Space Communications (G.M. Tosi Beleffi)

Telegraphy is a word coming from ancient greek and means in Italian scrivere a distanzawhile in English sounds more or less like writing to a distant place. The human being hasfrom the very beginning tried to increase his capabilities to communicate with his far awayfellow men and so to transmit. Under this point of view, the mythology is full of interestingexamples with the most famous and known that is Ermes, the Gods messenger, able to movefaster than the wind and responsible to carry informations to the Gods.First experiences in the ancient past can be found in the IVth century b.C. (before Christ),where Diodoro Crono reports on a human chain used by the Persian king Dario I (522-486b.C) to transmit informations from the Capital to the Empires districts.In the IVth Century b.C., Enea il Tattico, reports on an hydraulic telegraph probably

invented by the Chartaginians. During the Roman and Greek age, was used to place ingeographical key points fire towers to be switched on in case of security breachs and/orattacks on the borders. Eschilo (525-456 a.C.) reports in the Orestea that the news about thefalls of Troy arrived to Argo passing through the Cicladi islands covering, more or less, 900km (Eschilo, 458 b.C.). This sort of tradition remained, for example, on the Italian territoryassuming and adopting different schemes, fire or mechanical systems, depending on thetime period, the geography and the geopolicy (Pottino, 1976).In the Center-South of Italy, in particular, the use of fire based signals during night and ofsmoke based signals during the daylight on the top of towers or hills, afterwards calledcommunications by the usage offani, has been quite common in the XVI and XVII Centuries

www.intechopen.com


3/41


a.C. (Agnello, 1963). During the day one smoke signal means the presence of one enemyvessel, in the night was switched on a bundle of dry woods and moved up and down toinform about the exact number of the enemy vessels. Several testimonies report on differentcommunications links and distances. The most interesting one has been established in 1657

between the city of Messina and the Malta island with mid span vessels used to cover theMediterranean sea (Castelli, 1700).The use of mechanical systems to implement optical wireless systems is due to ClaudeChappe in 1792 (Huurdeman, 2003). Chappe introduces the optical telegraph in France.The system was based on a regulator, 4.5m long and 0.35m wide, to which two indicatorswere attached. This systems was placed on the top of stations in LOS (Line Of Site) at 9 kmeach. Telescopes and human repeaters were, of course, needed to move the regulator andthe indicators via three cranks and wire ropes. The time usage was short because the systemwas able to work only during the daylight and with good weather conditions. On the otherhand, it was long reach considering an average coverage in France equal to around 4830km,with 29 cities connected using around 540 towers. Security was ensured by transmittingsecret codes with short preambles, this also to understand the accuracy of the transmission.Chappe introduced, infact, a particular code in 1795, to increase the transmission speed. Thissystem helped to reduced the time to exchange informations from several days to minutesand has been adopted in 1794. The subsequent studies on the electricity, the results fromVolta (1745-1827) and from Ampere (1775-1836) on the electrical pile and the introduction ofthe electrical telegraph in 1838 (Morse), will carry to the dismission of the Chappe systemaround the mid of 1800. The Chappe system was introduced also in other Europeancountries connecting the cities of Amsterdam, Strasbourg, Turin, Milan and Brussels.At the end of the 19th century, Alexander Graham Bell experienced with excellent resultsthe so called Photophone (Michaelis, 1965) (Bova and Rudnicki, 2001). This system worked

using the sound waves of the voice to move a mirror, responsible to send pulses of reflectedsunlight to the receiving instrument. In particular Bell modulated with his voice, by the useof an acousto-optic transducer, a lens-collimated solar beam. Bell used to consider thisinvention to be his best work, even better than his demonstration of the telephone.Although Bells Photophone never became a commercial reality, it demonstrated the basicprinciple of optical communications.Wireless Optical Communications, becomes from this point and year by year moreimportant boosting the research worldwide. We can in this case divide the wireless opticalexperiments in three main areas depending on the time periods: in the 60s arrives the laserconcepts and rises up the idea of wireless communications, in the 90s becomes popular theidea of ground to satellite and satellite to ground laser communications still using red and

green sources, after 2000 the explosion of the Free Space Optical technologies (FSO) facescivil and military applications ranging from standard telecommunications up tointersatellites & interplanets experiments and using different wavelengths from 1 up to 10microns.For these reasons, essentially all of the engineering of todays FSO communications systems,has been studied over the past 40 years, at the beginning for defense applications andafterwards for civil ones. By addressing the principal engineering challenges FSO, thisaerospace/defense activity established a strong foundation upon which todays commercialFSO systems.

www.intechopen.com


4/41


In particular, the realization of the first LASER, based on ruby, in 1960 by Maiman openedwider possibilities for the communications involving beams propagating over longdistances in atmosphere. Low loss optical fibers (less than 20 dB/km), infact, will arrive onlyin the 70s. In 1960s NASA started to perform preliminary experiments between the Goddard

Space Center and the Gemini 7 module. In 1968 the first experiment about FSO transmissionof 12 phone channels along 4km had been demonstrated in Rome (Italy) by researchers fromthe Istituto P.T, CNR and Fondazione Ugo Bordoni under the management of Prof. Sette,Phisic Insitute University La Sapienza. A red laser source (0.8 microns) was used to connecttwo buildings between the Colombo and Trastevere Streets in Rome (Unknown, 1968). Inthe same year Dr. E. Kube in Germany published on the viability of free space opticalcommunications considering both green (0.6 microns) and red (0.8 microns) laser sources(Kube, 1968). The introduction of semiconductor light sources working at roomtemperature, by Alferov in 1970, were decisive for a further development of integrated andlow cost FSO systems. On the point of view of the research, the first experiment using aquantum cascade laser (Capasso 1994) can be considered fundamental today speaking aboutnew transmission wavelengths for FSO (up to 10 microns). Between 1994 and 1996 years thefirst demonstration of a bidirectional space to ground laser link between the ETS-VI satelliteand the Communications Research Laboratory (CRL) in Koganey (Tokio) has beenaccomplished. 1Mbps using 0.5 microns and 0.8 microns emitting lasers. With the ongoingintensive and worldwide studies on FSO communications, especially re started after theSeptember 11 tragedy where the communications were supported by free space optics links,the related scenarios changed extremely fast covering today different applications andenvironments like the followings: atmosphere, undersea, inter satellites, deep space. We caninfact report on the SILEX experiment (Semiconductor Intersatellite Link Experiment) in2001 demonstrating bidirectional GEO-LEO and GEO-ground communications. ARTEMIS

satellite (GEO) using a semiconductor laser at 0.8 microns directly driven at 2 Mbps with anaverage output of 10mW towards a Si-APD on SPOT-4 satellite (LEO). In the same year, theGeoLite (Geosyncronous Lightweight Technology Experiment) experiment successfullydemonstrated a bidirectional laser communication between GEO satellites, ground andaircraft. We cannot forget afterwards the MLCD (Mars Laser CommunicationDemonstration) program started in 2003 and ended in 2005 with the aim of covering thedistance between Earth and Mars planets using an optical parametric amplifier with anaverage output of 5W and photon counting detectors working at 1.06 microns (Majumdarand Ricklin, 2008).

3. Basic principles of the optical wireless communications(E. Leitgeb B. Geiger F. Nadeem - A.L.J. Teixeira, P. Andre)

3.1 IntroductionFree Space Optical communication links transmit information by laser light through anatmospheric channel. Relying on infrared light, these communication systems are immuneto electromagnetic interference (EMI), jamming, or wiretapping. Furthermore, they do notcause EMI themselves and operate at frequency bands (around 300 THz) were the spectrumis unlicensed. As a further advantage, FSO and fiber equipment can be combined withoutintermediate conversion, since both the air and the material used for fiber cables have goodtransmittance at the established wavelengths, namely 850 nm and 1550 nm. Currently, all-

www.intechopen.com


5/41


optical fiber/FSO systems are a well populated field of research, developing solutions forsignal regeneration, transmission, and reception without an intermediate electronic signal.FSO systems can be installed faster and cheaper than their wireless radio counterparts,making them interesting for short-term installations for events, military purposes and

disaster recovery. Consequently, a multitude of FSO equipment is commercially availablefor interconnection with standard fiber or Ethernet components. Acting as an alternative toother wireless radio or high-bandwidth wired links (fiber optics, Gigabit Ethernet), it has tofulfill general requirements such as low bit error rate (BER < 10-9) and high reliability.As already mentioned, a prerequisite for these requirements is an unobstructed line-of-sight,especially in long-distance outdoor environments. A major drawback therefore is thesusceptibility of FSO links to certain weather conditions, where especially fog causes severeattenuation of the laser beam and subsequent total link loss. Even moderate continental fogcan result in an attenuation of 130 dB/km, whereas dense maritime fog can account forattenuations up to 480 dB/km (E. Leitgeb et alt, 2006; M.S. Awan et alt, 2008).Rain attenuation has very little effects on the availability of FSO systems, but these effectsstrongly depend on the rain rate R. According to (T. Carbonneau and D. Wisley, 1998) andthe references therein an adequate relationship between rainfall and attenuation would be

5 2

exp 0 05556 0 00848

3 66 10

rain ( ( . . R

. R ) l)

(1)

For light to moderate rain rates of R = 5 mm/h as they are occurring in the continentalclimate of middle Europe the attenuation is only approx. 3 dB/km. Peak attenuations due totropical rain falls of R = 100 mm/h would result in higher attenuations (approx. 30 dB/km),but such weather conditions occur rarely and only in burst in Europe and the United States

(J. Wells, 2009; H.Alma and W. Al-Khateeb, 2008). Similar considerations hold for heavysnowfall (more than 5 cm over 3 hrs), where attenuations of more than 45 dB/km have beenmeasured (R. Nebuloni and C. Capsoni, 2008). Depending on seasonal and geographicpeculiarities, these values can vary to some extent. It may also happen that certain weatherevents occur simultaneously, i.e. heavy rain in combination with fog, or fog in combinationwith snowfall (V. Kvicera, 2008).Another phenomenon occurring influencing FSO communication links is related toscintillations and air turbulences. Air cells with different temperatures randomly distributedalong the link cause focusing and defocusing of the link due to changes in the refractiveindex. Amplitude and frequency of these scintillation depend on the size of cells comparedto the diameter of the optical beam (S. S. Muhammad, 2005; A. Chaman Motlagh, 2008). FSOsystems usually cope with such variations in the optical received signal strength (ORSS) byusing multiple beams (so-called multi-beam systems) or by using saturated amplifiers (M.Abtahi and L. Rusch, 2006). More detailed investigations can be found in (S. S. Muhammad,2005) and the references there.Other problems affecting visibility are mostly related to the narrow beam FSO systems use(usually at the order of a few milli radians): sand, dust, birds, et cetera flying through thebeam cause momentary link losses, whereas misalignment due to tower sway or thermaleffects can be coped by auto-tracking systems (J. Wells, 2009). The sun itself acts as a noisesource, which may completely overdrive the receivers (W. Kogler, 2003) if they are directlyexposed to sunlight. Soiling and aging of the components, especially lenses and mirrors,

www.intechopen.com


6/41


finalize the list of effects on FSO link availability. Summarizing, most of these effects can beovercome by either granting a certain link margin (snow and rain attenuation, fluctuationsdue to scintillation) or by adding complexity to the system (multi-beam and auto-trackingsystems). Fog, on the other hand, is the only remaining condition harmful for availability,

making carrier class availability (99.999%) for FSO systems highly questionable.Depending on the geographic areas, fog mainly occurs during fall and winter months on apersistent basis, whereas outages during summer and spring are caused by thunderstorms (E.Leitgeb, 2004). Fig. 1 shows the average unavailability throughout the year in Graz, Austria.Moreover, diurnal changes affect the probability of fog as Fig. 2 shows; it is low during noonwhere the sun clears up the sky and high during dusk, dawn and the night (E. Leitgeb, 2004).

Fig. 1. Average unavailability throughout the year (based on measurements from Oct. 2000

to Sep. 2001 (with the permission of E. Leitgeb, 2004))

Fig. 2. Probability of failure during the daytime (based on measurements from Dec. 2000(with the permission of E. Leitgeb, 2004))

Due to the complexity connected with phase or frequency modulation, current free-spaceoptical communication systems typically use intensity modulation with direct detection

www.intechopen.com


7/41


(IM/DD). Like in fibre optics communications systems, the performance characteristics for afree space optical communications system are dependent on the propagation medium.However, in this case the propagation medium is randomly changed, and susceptible toatmospheric conditions, resulting in alterations to the beam propagation constants.

In order to get a clearer picture of the behavior of FSO systems a numerical model for theatmosphere is going to be presented [Andre, 2003]. This propagation model wasincorporated in a commercial available photonic simulator and used to predict thebehaviour of a point to point free space data link as function of the climacteric variables.

3.2 Atmosphere ModelAs, referred, atmospheric effects can degrade free space data links by two mechanisms: (i)reduction in the detected optical power level due to atmospheric attenuation and (ii)random optical power fluctuations in the received beam which result in beam deformation,scintillation effects and beam wander (Kim, 1998). All these factors can become impairing to

the communications if their influence is significant. For that it is going to be presented eachof the contributing parts model and therefore a complete evaluation with effects will bemade for better understanding of the real effects in the system.

A. Atmospheric attenuationThe atmospheric attenuation results from the interaction of the laser beam with airmolecules and aerosols along the propagation. Similar to other waves, the optical beampower has an exponential decay with propagation distance. At a given distance from theemitter, l, the optical transmittance is:

exp0a s

P(l)

( l)P( ) (1)

where is the overall attenuation coefficient, determined by four individual processes:molecular absorption, molecular scattering, aerosol absorption and aerosol scattering.The molecular absorption includes the absorption by water, CO2 and ozone molecules. Theaerosol absorption results from the finely dispersed solid and liquid particles in theatmosphere, such as ice and dust, with a maximum radius of 20 m. A simple approach tocalculate absorption, assumes that variations in the transmission are caused by changes inthe water content of the air. The precipitable water, w (in millimetres), encountered by thelight beam is (Wichel, 1990):

310w l (2)

where is the absolute humidity in g/m3. This value can be related with the relativehumidity percentage (RH) and with the temperature in degrees Celsius, T, by:

).T/.

....T/..(RH

5213exp485

6713exp9690740

(3)

The absorptive transmittance can be then calculated for any transmission window, by thefollowing empirical expressions (Wichel, 1990):

www.intechopen.com


8/41


1 2exp /a i i A w , w w (4)

,

i

i

a i i

w k w w

w

(4a)

The values typical values of the constants used are listed in table 1 (these are also used incalculations following).

Windowboundaries(nm)

Ai ki i Wi

720 940 0.0305 0.800 0.112 54

940 1130 0.0363 0.765 0.134 54

1130 1380 0.1303 0.830 0.093 21380 1900 0.211 0.802 0.111 1.1

1900 2700 0.350 0.814 0.1035 0.35

2700 4300 0.373 0.827 0.095 0.26

4300 - 6000 0.598 0.784 0.122 0.165

Table 1. Constants used in expressions (4) and (4a).

Another attenuation process is the scattering, where there is no power loss, and there is onlya directional distribution. The two dominate scattering mechanisms are the Rayleighscattering, when the wavelength of the light is higher than the particle size, and the Miescattering when the particle size is comparable with the wavelength of the radiation. Anempirical relationship sometimes used to describe the scattering transmittance is [Wichel,1990]:

41 2exp s l C C (5)

where, C1 and are constants determined by the aerosol concentration and size distributionand C2 = 0.00258 m3 accounts the Rayleigh scattering. These two constants can be relatedwith the visual range, V, in kilometres at 550 nm [1]:

1

3 910 55

.C .

V (6)

For a very good visibility, can take a value of 1.6, and for average visibility it have a valueof 1.3. If the visual range is inferior to 6 km, them the exponent can be obtained by:

1 30 585 / . V (7)

The presence of precipitation (rain or fog) increases the scattering coefficient. Thetransmittance can be related with the rainfall rate (R) in mm/hr, by:

www.intechopen.com


9/41


5 2

exp 0 05556 0 00848

3 66 10

rain ( ( . . R

. R ) l)

(8)

The propagation of a laser beam in dense fog or clouds much difficult and attenuations ashigh as 50 150 dB/km can be found (Strickland, 1999).The total attenuation to be considered in (1) is the sum of the several partial attenuationfactor (eqs (4), (5) and (8))The geometrical beam expansion, resulting from the beam divergence, is also responsible fora reduction of the optical power coupled to the receiver. It must be also take into account theoptical miss alignment between the emitter and the receiver, for systems without auto-tracking (Kim, 1998).

B. TurbulenceThe atmospheric turbulence arises when air parcels at different temperatures are mixed by

wind and convection. This effect produces fluctuations in the density and therefore in the airrefractive index. The parameter that describe the disturbances caused by turbulence is therefractive index structure coefficient, Cn, which usually varies between 5 10-7 m-1/3 and 8 10-9 m-1/3 for situation of strong and weak turbulence, respectively.The value of Cn can be estimated by (Strohbehn, 1978):

6

24 3

2

4 3 13

2

53 10 3 16 2

79 10

273.15

3000 5.49 10

sin( )2 2 10 3000 1027

- /

n - /

- - - -

P

ThC =

v . e + e

(9)

where h is the height in meters, P the air pressure is milibars, v is the wind speed in m/s and the angle between the beam and the wind.The dominant turbulence scale size leads to different effects: (i) if the scale of the turbulencecells is larger than the beam diameter then the dominant effect is the beam wander, that isthe rapid displacement of the beam spot, (ii) if the scale of the turbulence cells is smallerthan the beam diameter then the dominant effect is the beam intensity fluctuation orscintillation.The radial variance of beam wander can be described by (Zhu, 2002):

1 32 2 31 90 2/

r n . C w l

(10)

where w is the spot size at the transmitter.The scintillation is described by a log-intensity distribution (Clifford, 1981), with a variancegiven by [Wichel, 1990]:

7 6

2 2 11 621 23

/

/

i n

. C l

(11)

www.intechopen.com


10/41


The effect of scintillation can be smoothed by spatial averaging using a width detector area,multiple apertures detector or a spatial diversity with several receivers or emitters (Kim,1997).The presence of atmospheric turbulence is also responsible for the beam spreading,

contributing to the beam divergence, which is given by [Wichel, 1990]:

1 5 6 5 8 62 01 / / /t na . C l (12)

C. Thermal BloomingThe molecular absorption by the air of the beam energy, will lead to a temperature gradientin the medium that induces density and index refraction changes. In the presence of air flow(wind) results in a density wavefront destruction which leads in a beam bender directed tothe air flow.The displacement of the beam at the receiver is (Strohbehn, 1978):

2

05 1 1

6 100 sin( )

I( ) (n ) l

wu=

P v

(13)

where is the ratio of specific heats (with a value of 1.4 in air), n is the refractive index of theair and I0 is the beam optical power at the transmitter.III. SimulationIn this subsection, and for sake of understanding of the modeling described above, a set ofsimulations is presented based on the atmospheric model described in the previous section.

This model was implemented through Matlab in a commercial available photonic simulator,VPI from Virtual Photonics .The free space optical communications system used as reference for these simulations hadthe following parameters. The propagation distance was 1000 m, oriented in a 158 heading.The optical power of the beam at the emitter was 40 mW and at a 780 nm wavelength, witha radius of 10 cm and a divergence of 1 mrad. The optical beam is modulated at 2.048 Mb/s(E1) with an optical extinction ratio of 15 dB. The use of this low bitrate allows us to latercompare these results with some experimental ones (Almeida, 2001). For the receiver wehave considered a photodiode based O-E converter with a 0.85 responsivity and a 1.4Mbit/s bandwidth pulse reformatting electrical filter, preceded by a 6 dB attenuator toaccount the miss alignment losses.

We obtained, for several values of temperature and relative humidity, the attenuation of for1 km path link. These results can be observed in figure 3.

www.intechopen.com


11/41


40 50 60 70 80 90 1005

10

15

20

25

30

35

Relative Humidity (%)

Temperature(C)

0.43750.54690.6563

0.87501.0941.3131.5311.7501.9692.1882.4062.6252.8443.0633.500

Fig. 3. Optical attenuation for one km path link as a function of the temperature and relativehumidity.

It is clear, from figure 3 that the effect of temperature and humidity in the losses, due toabsorption and scattering is important. The total attenuation vary from 0.5 dB/km to 3.5dB/km.The effect of rain fall can also be analyzed with this model. Keeping the values oftemperature and relative humidity constant, 25 C and 80 % respectively, the attenuationwas obtained as function of the rain fall rate, as displayed in figure 4.

1 2 3 4 5 6 7 8 9 10 11 12 13 141.0

1.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

1.9

Attenuation(db/km)

Rain Fall Rate (mm/h)

Fig. 4. Optical attenuation for 1 km path link due to rain. The line is a visual guide.

www.intechopen.com


12/41


The introduction of turbulence in the atmosphere model will result in the observation ofscintillation on the received power. In figure 5 is shown the BER (bit error rate) of thereceived data for 1 km path direct optical link as function of the received optical power andfor several values of the refractive index structure coefficient.

1E-7 1E-6 1E-51E-13

1E-12

1E-11

1E-10

1E-9

1E-8

1E-7

1E-6

1E-5

1E-4

1E-3

0.01

BER

Optical Power Received (W)

Cn = 5 x 10-8 m-1/3

Cn

= 1 x 10-8

m-1/3

Cn

= 5 x 10-9

m-1/3

Back to back

Fig. 5. BER versus the received optical power for several values of Cn. The lines are visualguides.

From the previous figure it is clear that the power penalty depends on the value of Cn, as it

can be observed in the eye diagrams of figure 6, obtained for a received power of 22 dBmand for several values of Cn. The eye diagram of figure 6 a) corresponds to a high turbulentmedium with a Cn value of 1 10-7 m-1/3, 6 b) is a situation of medium turbulence with Cnof 5 10-8 m-1/3, while 6 c) is obtained in a low turbulence medium with a value of 1 10-9m-1/3 for Cn.

a) b) c)Fig. 6. Eye diagrams obtained for several values of the refractive index structure coefficient:a) 1 10-7 m-1/3, b) 5 10-8 m-1/3, c) 1 10-9 m-1/3.

A reasonable good approach to estimate the BER of a FSO system is to consider only theattenuation (discarding the scattering and thermal blooming but considering the beamwander), then the BER can be written as:

www.intechopen.com


13/41


(14)

Where R is the detector responsivity, PR the optical power at the detector and the receiverthermal noise. The impact of the Cn factor in the BER can be observed in the figure 7, whereexperimental Cn factor measured in Rio de Janeiro along the day in February 2003. For thereceiver it was considered a typical configuration with R=0.9 A/W and a receiver diameterof 13 cm, the optical power at the emitter is 10 mW for a link length of 1 km.

Fig. 7. Cn influence in the BER for 1 km link.

4. Hybrid network infrastructures: FSO & RF(B. Geiger E. Leitgeb F. Nadeem)

4.1 IntroductionRelying on an unobstructed line-of-sight, FSO links are strongly influenced by atmosphericconditions reducing or influencing visibility, such as fog, precipitation, haze, andscintillation. Fog, as one can expect, is the most critical effect affecting attenuation and,subsequently, availability of the FSO link. Attenuation is caused by scattering, resultingfrom the fact that the size of the fog particles is in the order of the wavelength of optical andnear-infrared waves (as they are used for FSO). Consequently, link distances in coastal ormetropolitan environments which are prone to fog are limited to a few hundred meters.Radio Frequency (RF) links on the other hand show almost negligible fog attenuation if thecarrier frequency is chosen accordingly, while they usually suffer from other precipitationtypes like rain and wet snow. Combining these two technologies to an FSO/RF hybridnetwork may increase overall availability significantly, guaranteeing quality-of-service andbroadband connectivity regardless of atmospheric conditions.Several types of hybrid system concepts have been introduced in the literature (L. Stotts etalt, 2009; S. Bloom and W. S. Hartley; H. Izadpanah et alt, 2003; T. Elbatt et alt, 2001;F. Nadeem et alt, 2009; E. Leitgeb et alt, 2004; J. Pacheco de Carvalho et alt, 2008; S. Vangala

www.intechopen.com


14/41


et alt, 2007; A. Akbulut et alt, 2005; S. Gurumani et alt, 2008; J. Derenick et alt, 2005; T.Kamalakis et alt, 2005; W. Kogler et alt, 2003; F. Nadeem et alt, 2009; S.D. Milner et alt, 2004;O.I. Kim and E. Korevaar, 2001; H. Wu et alt, 2004; S. Vangala and H. Pishro-Nik, 2007),focusing on increasing availability, bandwidth efficiency, or minimizing system complexity.

Resulting from these focuses, hybrid systems can be categorized in three different groups:Redundant, load-balancing, and switch-over hybrids.

4.2 Description of the RF Communication Links

As already mentioned, FSO links are strongly influenced by fog attenuation andconsequently suffer from long periods of total link loss (E. Leitgeb, 2005). On the other hand,FSO systems provide very high data rates without the requirement of licensing. As aconsequence, the RF link for a hybrid FSO/RF system has to be chosen according to thefollowing requirements:

The RF link should be available whenever the FSO link is not, i.e. it should not beinfluenced by fog or other weather effects reducing visibility. The RF link should provide a similar bandwidth as the FSO link, so that the hybrid

system does not suffer from performance degradation. The RF link should be operated in a frequency band which does not require

licensing, so that this advantage of FSO systems is not lost in a hybrid setup.

Unfortunately, some of these requirements are contradictory. High data rates, orequivalently, bandwidths require high carrier frequencies, which on the other hand eithershow strong attenuation due to fog or would result in a prohibitively high systemcomplexity. Moreover, these systems operate in license-free, but regulated bands and are

thus subject to stringent transmission power restrictions limiting the possible link marginconsiderably. A higher geometrical loss further adds to the availability issues. Especiallyduring times when rain and fog occur simultaneously, as it often happens in continentalclimate, both links are suffering from weather effects (E. Leitgeb et alt, 2004; E. Leitgeb et alt,2005). Scintillations have little or no effect on RF links, since they are not susceptible tochanges in the refractive index rather than changes in humidity (S.S. Muhammad, 2005). Amore complete discussion of weather effects on RF links is available in (H. Wu et alt, 2004;ITU-R, 2005). Commercially available bands with very high available bandwidths arecentered around 60 GHz and 70/80 GHz, respectively. While the former, license-free bandcannot be exploited for long link distances due to an oxygen absorption peak (15 dB/km),the latter provides an interesting field of research for hybrid systems. Peak attenuation dueto moderate rain are usually well below 5 dB/km, whereas fog attenuation is as small as 0.4dB/km for a fog density of 0.1 g/m3 under these conditions a FSO system would sufferfrom 225 dB/km (S. Bloom, 2005). Currently available equipment operating in the 70/80GHz band can provide carrier class availability disregarding weather conditions over adistance of 2-3 km achieving 1 Gbps. Unfortunately, this band can only be used afterobtaining a low-cost license. Spectra currently unregulated by the ITU lying at 275 GHz areunreachable with current technologies (J. Wells, 2009).Lower frequency bands (e.g. the license-free ISM bands at 2.4 and 5 GHz), on the otherhand, provide much less bandwidth to the user, leading to a greatly decreased bandwidthperformance of the overall hybrid system. However, these systems are not susceptible to fog

www.intechopen.com


15/41


at all (ITU-R, 2005) and also show much smaller influence of rain and snow than systemsoperating with carrier frequencies beyond 20 GHz (H. Wu et alt, 2004). Multi-userinterference, as it is common in license-free bands at lower frequencies, which are accessibleby low-cost technologies, can be mitigated by using directional links.

These bands, which also host users of IEEE 802.11a/b/g/n compliant equipment, are ofparticular interest since they allow building a hybrid FSO/RF system with commerciallyavailable equipment for the RF part as well. Especially the IEEE 802.11a standard isinteresting for FSO/RF hybrids, since it is operating in the less populated 5 GHz band andallows a higher transmit power of up to 30 dBm EIRP. Although the standard claimspossible data rates of up to 54 Mbps, usual limits for long-range directional links are around20 Mbps. Moreover, the use of this technology offers the possibility of an upgrade to IEEE802.11n compliant equipment, claiming data rates of up to 600 Mbps with MIMOfunctionality. Future studies will show if this technology can also be used for long-ragedirectional links and thus build the RF component of the proposed hybrid systems.

4.3 Overview of Hybrid Systems introduced in the literature

Hybrid systems can be classified into three categories: Redundant systems: These systems duplicate data and transmit it simultaneously

over both the RF and the FSO link. As a consequence, the data rate of both links hasto be equal, resulting in either a requirement for very high frequencies on the RFlink or a relatively high FSO underutilization. Moreover, systems which duplicateand recombine data are necessary. Redundant systems provide a high reliability,but suffer from the fact that both links have to be active all the time, wasting asignificant amount of energy.

Switch-Over systems: These systems transmit data only over one link, which is

chosen according to link availability. Usually, since the FSO link allows higher datarates, it is chosen as a primary link whereas the RF link acts a backup.Consequently, data rates of both links need not be identical, if one accepts areduced bandwidth during fog events. Switch-over systems require multiplexerson both ends, algorithms choosing the active link, synchronization, and accurate,timely measurement data of the optical signal strength. However, these algorithmssave energy by transmitting over one link only, and can be connected to standardnetwork equipment without protocol overhead (F. Nadeem et alt, 2009).

Load-Balancing systems: These most sophisticated algorithms distribute trafficamong the links according to the quality of their connectivity, thus exploiting the fullavailable bandwidth each time. Besides a measurement of the link quality, thesesystems require recombination systems on either sides of the hybrid link, oftenresulting in either significant protocol overhead (F. Nadeem et alt, 2009) or high-complexity codes which automatically distribute data among different links.

In the literature a wide field of hybrid systems can be found: AirFiber (S. Bloom, 2009), a US-based company pioneered redundant transmission over FSO and RF links, the latter onebeing a millimeter wave (MMW) link with a carrier frequency at 60 GHz. Data rates ofapprox. 100 Mbps were achieved, but availability was well below the expectations. Wu et alt.(H. Wu et alt, 2004) analyzed FSO and RF link separately and concluded that by using ahybrid network link margin can be reduced significantly to achieve carrier class availability.

www.intechopen.com


16/41


The FSO link was using a 1550 nm laser source, the RF link was operated in the 60 GHzMMW band. Link distance was 1500 m.

Fig. 8. Redundant hybrid system and availability measurements (with the permission ofE. Leitgeb et alt, 2004)

Kim and Korevar (I.I. Kim and E. Korevaar, 2001) studied the distance limitation of FSOsystems for both carrier and enterprise applications and showed that carrier classavailability can be achieved for much longer link distances if the FSO link is combined withan RF back-up transmitting data redundantly. Leitgeb et alt. performed an experiment inGraz, Austria, over 15 months during 2002-2003, where data was sent simultaneously overtwo FSO/RF links (E. Leitgeb et alt 2004). The RF link was designed with a carrier frequency

of 40 GHz, the FSO system transmitted using a 850 nm laser with a data rate of 155 Mbps.The availability was measured over this time for each individual link and for the hybridcombination. The redundant transmission achieved an availability of 99.93% (see Fig. 8.)Hashmi et alt. (S. Hashmi and H. Mouftah, 2004) also proposed a redundant hybrid system,calculating it based on rain data only for an FSO and a MMW link in the 60 GHz band. Theyalso mentioned that the hybrid system could be used in an asymmetric uplink/downlinkscenario where different traffic demands have to be served.Akbulut et al (A. Akbulut et alt 2005) developed an experimental hybrid FSO/RF switch-over system between the two of five campuses of Ankara University, Turkey, located atdifferent locations in the city. The optical link provided a 155 Mbps full duplex connectionby using a laser source at 1550 nm over a distance of 2.9 km. The RF link was compliant toIEEE 802.11b WLAN, operated at 2.4 GHz linking the two terminals at 11 Mbps. The switch-over algorithm was a power hysteresis. Pacheco de Carvalho et alt. (J. Pacheco de Carvalho,2008) installed a similar system at the University of Aveiro, Portugal, where a 1 Gbps laserlink was backed up by a 75 Mbps (nominal) WiMAX (IEEE 802.16) link. The laser link wasoperated at 1550 nm, switching was implemented on the network layer via switchingbetween static routes. Power hysteresis was employed, and the link distance was 1.14 km.Milner and Davis (S.D.Milner and C.C. Davis, 2004) proposed a switch-over system fortactical operations in a general manner, considering protocols for switching between links aswell as for traffic re-distribution after a change in the network topology. Their intention wasto use two 1550 nm FSO systems in combination with an RF link operated in the Ku-band

www.intechopen.com


17/41


(12-18 GHz). Kamalakis et alt. (T. Kamalakis et alt, 2005) installed a hybrid switch-oversystem at the University of Athens, Greece, using a 1 Gbps FSO link and a 100 Mbps MMWlink operated at 95 GHz over a distance of 800 m. Also, (L. Stotts et alt, 2009) report aboutswitch-over systems.

Dynamic load balancing is also proposed in literature: ElBatt and Izadpanah ( H. Izadpanahet alt, 2003; T. ElBatt and H. Izadpanah, 2001) proposed a load balancing system distributingtraffic among one FSO and several RF links. In this work, however, it is assumed that theamount of traffic on the FSO link affects it availability. Vangala and Pishro-Nik (S. Vangalaand H. Pishro-Nik, 2007; S. Vangala and H. Pishro-Nik, 2007) use special non-uniform low-density parity check codes to distribute traffic among different links, showing increased linkutilization and availability, while BER could be reduced significantly. Finally, Nadeem et alt(F. Nadeem et alt, 2009; F. Nadeem et alt, 2009) analyzed both switch-over and load-balancing systems based on standard Ethernet equipment with minimum hardwareextension. A 155 Mbps FSO system with a 850 nm laser was used in combination with anIEEE 802.11a link. It was shown that availability almost achieves carrier class values of99.999%.

4.4 Hybrid FSO/RF switch-over systemSwitch-over (SO) systems, as they were introduced in Section IV can be illustrated by Fig. 9:Depending on the strength and availability of the links, only one of them is used fortransmission. While the FSO link is the primary link, the RF link acts as a back-up. In thissection the interested reader will find an overview of the problems in designing such an SOsystem together with some possible solutions. In particular, synchronization between theswitches/multiplexers on both sides, SO algorithms and possible applications will beanalyzed.

Fig. 9. Application setup of FSO-WLAN switch-over system (with the permission of(F. Nadeem et alt, 2009)

For the simulations, a commercial WLAN link with a carrier frequency in the 5 GHz ISMband will be considered. The WLAN link was built with two embedded PCs using high-gaingrid antennas and miniPCI WLAN cards with fairly high receiver sensitivity (depending onthe antennas, distances of over 50 km can be covered). The FSO system is a GoC MultiLink155/2 system. It supports data rates of 155 Mbps over distances up to 2 km and uses 4transmitters at 850 nm. The properties of the FSO and the WLAN system are given inTable 2. It is further assumed that the transceivers of both links are able to provide statusinformation.

www.intechopen.com


18/41


19/41


loss (J. Pacheco de Carvalho et alt, 2008). In the former case the router needs special hardwareextensions or interfaces to some measurement equipment, in the latter the router has to becapable of running a custom-made program. Furthermore, switching based on physical signalstrengths has the advantage that it does not rely on actual errors, but allows for

implementation of a more or less generous link margin. Switching itself is usually done bychanging the route metrics (J. Pacheco de Carvalho et alt, 2008) of the different links accordingto the measurements. Consequently, all packets are transmitted, although not all may reach itsdestination. Moreover, one can expect that depending on the type of link status informationavailable reactions on changes in the ORSS are delayed. Links switched on the network layerare usually not transparent to lower layers. In order to provide full connectivity,synchronization between the different multiplexers has to be ensured. Assuming that thechannel is reciprocal one may state that both sides of the system will always measure the sameORSS and therefore chose the same link as active automatically. However, usually one doesnot want to base a hybrid system designed to achieve maximum reliability on that assumption.Moreover, using a side channel for transmitting information about which link to take isquestionable as well, because that very channel has to be made reliable itself. Of course, onecan feed such information into both the FSO and the RF link, but that certainly adds to thecomplexity of the system. Besides, it was the main intention of the SO system that at each timeinstant only one link has to be active. Finally, an asymmetric scenario, where one of themultiplexers chooses the link according to the ORSS would be possible. The other multiplexeraccepts packets from either link, but responds only over the very link from where the lastpacket arrived. Such a self-synchronizing setup, as it was introduced in (F. Nadeem, 2009) hasthe disadvantage that it inherently relies on quasi-continuous transmission from the networkon the side of the active multiplexer a condition which is usually fulfilled by higher-layerprotocols, such as TCP. In any case, despite all considerations about synchronization and

switching, the overall hybrid network still has to be considered unreliable; the difference,however, is that the availability is increased significantly.

Fig. 10. Comparison of discrete and continuous ORSS values. (with the permission ofB. Flecker, 2006) Fog event from October 25th, 2005, 03:00 to 11:00.

www.intechopen.com


20/41


Switch-Over Algorithms: After deciding upon the layer on which switching is performed,upon the data which is determining the active link, and the synchronization method, thedesigner of the hybrid SO system has to take a multitude of switch-over algorithms intoconsideration.

Remembering the advantages of physical measurements describing the link status we willput special emphasis on SO systems switching based on those measurements. This,however, rises an important question: Can it be guaranteed that the FSO system (whichpurely determines the active link) provides link status information in an accurate, timelymanner? It can be taken as granted that the system itself measures and uses ORSSinformation, but is this information accessible to the user over a certain interface? TheMultiLink 155/2 system for example indicates the ORSS continuously with a LED bar(discrete amplitude as shown in Fig. 10), but provides this data over an RS232 interface on aper-second basis only (discrete time). Somewhere within the system, however, the time andamplitude continuous ORSS will be available (as also shown in Fig. 10). Thereby one canassume that the previous assumption of link status information provided to the user can bejustified, even if hardware reconfiguration are necessary. As one can see in Fig. 10,especially during the gradient from clear sky conditions to foggy weather there are manyvariations in the ORSS. These variations, as one can expect, cause a certain threshold to becrossed multiple times. If now the SO system is designed to employ a straightforwardthreshold comparison (TC) algorithm, it would suffer from frequent switching between thelinks due to these variations. Since after each switching operation a certain time is requiredto restore the link completely (so called link loss time, or LLT), frequent switching wouldcause reduced bandwidth and availability. Consequently, other algorithms coping withthese variations have to be evaluated it is the purpose of this section to introduce some ofthem and compare their performance.

a) Power Hysteresis (PH): A power hysteresis defines two thresholds and two states: a lowerand an upper threshold, WLAN and FSO operation. If during FSO operation the lowerthreshold is crossed, WLAN is activated. If during WLAN operation the upper threshold iscrossed, FSO is activated its as simple as that to prevent the system from switching backand forth. The width of the hysteresis (i.e. the distance between upper and lower threshold)depends on the amplitude of variations and has to be optimized with respect to actualsystem measurements. To maximize availability, the lower threshold has to be set to valuesequal to or greater than the receiver sensitivity of the FSO device. b) Time Hysteresis (TH):Relating variations in the ORSS to bouncing of electrical contacts, one can also usetechniques called debouncing to cope with these variations. Such techniques usually employa wait period T during which the ORSS are evaluated and during which after every

threshold crossing the wait period is restarted. Consequently, only if the signal does notcross the threshold for a certain time, an SO operation is performed. The duration of thewait period in that case is determined by the frequency and the amplitude of the variations.To maximize availability, the wait period can be set to different values for crossing thethreshold in either directions; in the limiting case, the wait period can even be omitted forswitching from FSO to WLAN. c) Filtering: Treating variations in the ORSS as noise,methods for noise mitigation come into view. Most prominently, low-pass filtering can benamed as such a method. Different realizations of low-pass filters in the analog (RC-network) and digital (moving average filter, raised-cosine filter, etc.) domain arepossibilities to cope with this unwanted noise. Filters are characterized by their order, pass

www.intechopen.com


21/41


and stop band characteristics, and by their cut-off frequency. These characteristics have to bedesigned with respect to the frequency of the variations. It is of vital importance that thefrequency of these variations is in the stop band, while the highest occurring frequency inclimatic changes still lies in the pass band to allow for a timely reaction on an emerging fog

event.Moreover, combinations of these methods can be considered (e.g. filtering and hysteresismethods, power hysteresis and debouncing, etc.). Unfortunately, seasonal and diurnal, aswell as geographic peculiarities make a general design or general optimization ofparameters impossible. The reader will understand that the design of a SO system in acoastal area with dense fog conditions differs from a system in a metropolitan area, wheremoderate fog can be expected. The following simulations therefore focus on the continental,metropolitan climate of Graz, which is characterized by moderate, persistent fog eventsduring fall and winter and strong rainfall during summer.Simulations and Results: For the simulation, measurement data was taken from (B. Fleckeret alt, 2006). For parameter optimization the particular fog event depicted in Fig. 10 wasused. The benefit of focusing on one fog event is based on the fact that the influence of LLTafter switching is increased compared to the influence on an all-year average availability.Moreover, as it can be seen in Fig. 1, fog events mainly occur during fall and winter, makinga separate analysis of these seasons sensible. Assuming little or no unavailability during thesummer months, simulation data can be extrapolated. Receiver sensitivity of the FSO systemwas set to -22 dBm. For this value, a significant number of threshold crossings occurredwhich allows an optimization of the algorithms. Using this sensitivity, the fog event underconsideration yielded an FSO availability of only 67.43%. Link loss time after switching wasset to 3 s in order to include an additional margin to link re-establishment. For bandwidthsimulations, bandwidths of the FSO and WLAN link were set in accordance to (F. Nadeem

et alt, 2004) to 91.9 Mbps and 18.8 Mbps, respectively. The WLAN link was assumed to beactive whenever FSO was inactive. This assumption holds for fall and winter periods whereFSO outages are usually caused by fog only and where rain is rarely occurringsimultaneously. During summer months where strong rainfall in combination with severefog affects both links this assumption may not be valid anymore (E. Leitgeb et alt, 2004).As one can imagine, finding the best algorithm parameters is related to finding a trade-offbetween availability and bandwidth efficiency. While WLAN may be available throughoutthe year, its bandwidth is prohibitively low. Consequently, the simulations are limited by aminimum bandwidth of 60 Mbps. A more complete evaluation of simulation results and acomprehensive discussion of this topic can be found in (F. Nadeem et alt, 2009).a) Threshold Comparison: Pure threshold comparison (TC) is done by comparing the ORSS

to the RX sensitivity and switching based on the outcome of this comparison. TC yields anincrease in availability to 98.62% while achieving best bandwidth performance (see Table 3).This can be explained by the fact that TC uses the FSO link whenever it is available, and theoutstanding bandwidth of this link compensates for relatively high unavailabilities due toLLTs. However, for maximizing availability this may not be the best of all choices as Fig. 11shows. b) Power Hysteresis: For all simulations, the width of the power hysteresis was set to1 dB and the lower threshold was varied. Fig. 12 shows that availability can be increasedsignificantly by increasing the lower threshold to values much greater than the receiversensitivity. The only problem is that by increasing this threshold FSO underutilizationincreases and, subsequently, bandwidth efficiency is low. E.g. to obtain a minimum

www.intechopen.com


22/41


bandwidth of 60 Mbps, the lower threshold has to be below -21.7 dBm. In these regions, alsothe beneficial effects of filtering cannot be exploited anymore, because the time to react onchanges in the ORSS due to fog increases.

Fig. 11. Bandwidth for different switch-over methods (with the permission of F. Nadeem etalt, IET submitted 2009)

0 0.5 1 1.5 20.99

0.991

0.992

0.993

0.994

0.995

0.996

0.997

0.998

0.999

1

Lower Threshold above RX sensitivity in dBm

A

vailability

Pure PH

MAPH (10th order)

MAPH (20th order)

MAPH (40th order)

MAPH (60th order)

Fig. 12. Availability for pure PH and MA-PH (with the permission of F. Nadeem et alt, 2009)

www.intechopen.com


23/41


0 10 20 30 40 50 600.975

0.98

0.985

0.99

0.995

1

Wait period T in s

Availability

Pure TH

MATH (10th order)

MATH (20th order)

MATH (40th order)MATH (60th order)

Fig. 13. Availability for pure TH and MA-TH (with the permission of F. Nadeem et alt, 2009)

0 0.5 1 1.5 20.98

0.982

0.984

0.986

0.988

0.99

0.992

0.994

0.996

0.998

1

Threshold above RX sensitivity in dBm

Availability

MA (10th order)

MA (20th order))

MA (40th order)

MA (60th order)

Fig. 14. Availability for different filter orders (with the permission of F. Nadeem et alt, 2009)

c) Time Hysteresis: For the time hysteresis the threshold was set to the receiver sensitivity,and the wait period T was varied as a simulation parameter. As seen in Fig. 13, availabilitycan be increased significantly. Moreover, FSO underutilization is low, so the minimumbandwidth of 60 Mbps is achieved for all depicted values of T. Filtering beforehand is

www.intechopen.com


24/41


counter-productive. d) Filtering: For filtering, the threshold was increased in steps startingfrom the receiver sensitivity. As the only filter type, a moving average (MA) filter wasconsidered, where the order N of the filter automatically determines its cut-off frequency(other filter types are discussed in (F. Nadeem et alt, 2009). Fig. 14 shows that availability

increases with increasing thresholds. Interesting, though, might be the fact that higherorders (i.e. lower cut-off frequencies) perform better than lower ones, as long as thethreshold is set to values high enough. High-order filters perform smoothing, but do notallow timely reactions on critical changes in the ORSS. Consequently, high availability isonly achievable with a combination of smoothing and a large margin to the receiversensitivity.This in turn leads to FSO underutilization and limits maximum threshold values to -21 dBmto obtain a minimum bandwidth of 60 Mbps. In these regions, however, lower filter ordersoutperform higher orders.e) Combined Power and Time Hysteresis: For the combined power and time hysteresis thelower threshold was set to the receiver sensitivity and the width of the hysteresis was 1 dB.The wait period of the time hysteresis portion was varied. As it can be seen in Fig. 15 andTable 3, pure PT delivers best results in term of availability. Furthermore, minimumbandwidth of 60 Mbps can be achieved for wait periods below 40 s, where availability stillhas values above 99.8%. Extending this simulation to the whole measurement campaign,availabilities of 99.988% can be achieved, as it is shown in Table 3. Simulations proved thatby doubling this period to T = 80 s, availability could be increased to 99.997%. Taking thesevalues into consideration one can see that carrier class availability becomes a graspable goal,even for hybrid switch-over systems.

0 10 20 30 40 50 600.988

0.99

0.992

0.994

0.996

0.998

1

Wait period in s

Availability

Pure PTMAPT (10th order)

MAPT (20th order)

MAPT (40th order)

MAPT (60th order)

Fig. 15. Availability for pure PT and MA-PT (with the permission of F. Nadeem et alt, 2009)

www.intechopen.com


25/41


Table 3. Performance comparison of different switch over methods (with the permission ofF. Nadeem et alt IET submitted 2009)

Fig. 16. Map of the campus of the Technical University of Graz

An Application: Interconnection of different sites of the campus Finally, to concludeabout hybrid switch-over systems, a possible application scenario shall be introduced,where different sites of the campus of Graz University of Technology will be interconnected.Such an application is widely evaluated in the literature (J. Pacheco de Carvalho et alt, 2008;A. Akbulut et alt, 2005), but only (F. Nadeem et alt, 2009) considers not only availability ofthe different links but also traffic demands of the different sites. In Fig. 16 one can see the

www.intechopen.com


26/41


location of the main sites of the campus, which are known as Alte Technik, NeueTechnik and Inffeldgrnde. The latter one is the largest, housing many offices andstudent computer rooms. Moreover, one can see that the distance between the sites neverexceeds 2 km, making the use of directional WLAN links and FSO links possible. For

evaluation purposes, characteristics described in Table 2 were considered. Moreover, it wasassumed that the line-of-sight for the FSO and for the WLAN link is free. As alreadymentioned availability of the FSO link is generally high during the summer months, andduring the winter months only during daytime (cf. Fig. 1 and Fig. 2) for the continentalclimate of Graz. Traffic demands were recorded using the Multi-Router-Traffic-Grapher(MRTG), where green bars indicate incoming and blue lines indicate outgoing traffic.

Fig. 17. Traffic data recorded for Inffeldgasse (with the permission of F. Nadeem et alt, 2009)

Traffic recordings were made during December 2008, assuming that the average availabilitydue to fog is similar as in Fig. 1. Fig. 17 shows the traffic demands for campus Inffeldgasseon December 3rd, 2008. It can be seen that peak traffic demands are occurring between 10am and 4 pm, medium traffic was caused from 8 am to 10 am and from 4 pm to 6 pm,whereas traffic during the night time is low. Obviously, the major traffic requirementscoincide with office and lecture hours. These considerations do not only hold for a particularday, but throughout the year naturally, on holidays and weekends, traffic demands are

much lower. Moreover, one can see that this peak of incoming traffic at 8 pm occurs everyday, which is most likely related to an automatic backup. Scheduling such events moreproperly, traffic demands can be distributed accordingly. Simulations were performed usinga set of measurements of the years 2000 and 2001 (J. Tanczos, 2002). Link bandwidth was setto 155 Mbps for FSO and to 15 Mbps for WLAN, respectively. For the WLAN link a slightlylower bandwidth was taken, assuming that the Fresnel zones may be partially blocked bysurrounding buildings and trees. Link loss time was neglected, since it affects the averagebandwidth only very little. A more complete discussion of these things can be found in(F. Nadeem et alt, 2009). Comparing Fig. 18 with Fig. 17, the diurnal changes in the trafficrequirements are reflected in the average as well. Furthermore, one can see from Fig. 18 that

www.intechopen.com


27/41


the proposed hybrid FSO/WLAN switch-over system can satisfy traffic demands onaverage. Unfortunately, peak traffic demands of campus Inffeldgasse exceed even theavailable bandwidth for FSO, let alone WLAN. However, by using multiple links of eachtechnology, or newer, more sophisticated equipment (such as Gigabit FSO equipment and

IEEE 802.11n standard compliant WLAN links), traffic demands could be satisfied in ahighly reliable manner over wireless links.

Fig. 18. Average required and achieved bandwidth (with the permission of F. Nadeem et alt,2009)

5. Blue sky applications: inter satellite, inter planetary, under sea, chip tochip FSO communications (G. Incerti G.M. Tosi Beleffi)

The today increase of the networks complexity, with several devices and subsystemsintensively used, involves an aggressive use of the bandwidth management thus toguarantee an high rate and security, especially in military scenarios. In fact, the militaryapplications requires more strictly features respect to the civilian applications. Thebandwidth offered by the optical cables is very high and for this reason the optical fibres arealso used in military area. Inside airplanes, UAVs, vessels, cars and so on. Severalinformations and data can run through the same optical fibre and an high rate can betransmitted and managed. We started with this introduction on military purposes becausethis is the first market that boost the optical wireless, from the paper to the realimplementation. FSO communication, infact, is a valid solution especially in militarysituations because of the previously mentioned ability to guarantee a confidentialtransmission with a huge bandwidth.

www.intechopen.com


28/41


Challenge Mitigation

approach

Description

Pointing,

acquisition, andtracking

RF/FSO hybrids

Adaptive optics

RF system facilitates coarse acquisition and trackingAn RF channel can serve as control channel for FSO data

linkAdaptive optics systems achieve very fine beam steeringand tracking

Weather/

environment

Path redundancy

RF/FSO hybrids

Path redundancy and topology control are implementedin an FSO network to counteract link obscurationEnvironmental obscuration for optical may bepermissible for RF or vice versa

Turbulence Adaptive optics

Channel

coding/Diversity

Adaptive optics correct beam distortionChannel coding/diversity improve BER through forwarderror correction

Eye safetyEye safety

Infrared wavelengthsAdaptive optics

Infrared wavelengths such as 1550 nm are more eye safe

than visible wavelengthsAdaptive optics reduce the need for increased power bycorrecting beam for improved SNR

Networking RF/FSO hybrids

QoS techniques

RF system provides channel for topology control, linkmonitoring, and broadcasting network statusDifferentiated services protocols sort data by priority tocounter capacity changesApplication layer QoS algorithms prioritize data

Table 4. FSO mitigation approach (Juan C. Juarez et al., 2006)

The limited scenario offered by the radio frequency (RF) spectrum available for military use,

contributed to the exploration of alternative systems able to convey the secret informationsgenerated by military devices and/or systems. RF based systems reach only hundreds ofMbps per link and the RF beam cover an high area, in terms of spatial aperture, thusincreasing the eavesdropping percentage. On the contrary, FSO systems can guaranteerobust optical link with a very small beam size. Granting, at the same time, a hugebandwidth in the order of Gbps. Confining the data flow in a small spatial portion representan advantage because becomes very difficult to detect the beam and subsequently dropsome information from one or more miscreants. Furthermore, several beams, close to eachother, can be used at the same time to transport the information without any kind ofinterference and or interaction.The precision in the pointing and tracking steps is still a challenge especially in complex and

variable scenarios like, for example, the sea one. Mounting FSO on vessels, infact, means,first of all, that a fast tracking system should be implemented. Maintaining, of course, aminimum power budget and the numeric aperture already set. In particular, for the militaryapplications, is required an high degree of accuracy to obtain an alignment of laser beamwith the receiver (Juan C. Juarez et al., 2006).In order to win the challenges induced by the adoption of FSO systems in thecommunication scenario, sometime is possible to discover the presence of RF backup lines,as supporting elements of the optical counterpart. This is commonly referred as an hybridcommunication system. The hybrid system shares a common aperture and the use of FSOwith RF system permit to facilitate principal function like for example control signalling,

www.intechopen.com


29/41


tracking, acquisition and signals discovery. The RF beam is used to search the other device(neighbour discovery) or to start the acquisition step. The RF beacon is used also as controlsignalling and to rehabilitate the communications in case of optical channels fades.In order to minimize the dependence from a single link, reconfigurable FSO links can be

accomplished. Path redundancy and topology control are two ways to set up a fast andsmart network able to counteract in case of path obscuration. Moreover, to compensate theenvironment effects without increase the total beam power, the adoption of adaptive opticssuch as deformable mirrors, is considered a proper solution. In the following figure, areillustrated several static and mobile nodes. This is a real scenario in which an FSO link canbe established in a few time.

Fig. 19. Topology of a battlefield scenario using FSO system to connect all partners (Juan C.Juarez et al., 2006)

Referring to the picture, node C has two optical heads and must be able to manage theoptical beams. Moreover, it is near at the B, D, E nodes; thus node C must be able to take adecision in which direction must point its optical beam. This kind of decision can bemanaged considering operational aspects; distance from the other devices, rate and trafficdemand, distance between the end users, and environment measurement because weatherconditions could limit the optical link. In this case, node C is able to send the beam throughanother path using another nodes. How shown in the picture, node D result isolated

because the dense fog does not allow to established a FSO link with the airborne node. Thus,the only way result node C but its optical heads are both in use. The idea of the opticalreconfigurable network is to establish a link between nodes C and D instead of node B.Node B will be reached from node.The quickness in term of time to install, together with a small size, make this kind oftechnology able to operate in different segments, like for example, the rehabilitation of linkin case of terroristic attacks or disaster recovery due to natural catastrophic events (E.Leitgeb et al., 2005).FSO technology is used also in several non conventional scenarios like:

www.intechopen.com


30/41


Aerospace communications: a laser beam can directionally guarantee a linkbetween two satellites, between the heart and a satellite platform, and vice versa,without any kind of interference and, at the same time, achieving an high data rate.Several experiments have been performed demonstrating that the FSO technology

is mature to accomplish such kind of challenges. On the inter satellite side. In 2008has been demonstrate an optical link between two LEO orbiting satellites, TerraSAR-X and NFIRE, at 5.5 Gbps on a total distance of 5500km and at a speed of25000 km/h. On the downstream side. The KIODO (KIrari's Optical Downlink toOberpfaffenhofen) project demonstrated a downlink stream from the OICETS LEOJapanese satellite. In 2006, 5 trials performed successfully achieving a BER of 10-6with a modulated optical signal at 50 Mbps and 847nm (N. Perlot et alt, 2007). TheFP6 CAPANINA (Communications from Aerial Platform Networks deliveringBroadband for ALL) project regarded the downlink between a stratospheric ballonat 22 km and a transportable ground station in Kiruna, Sweden at 1.25 Gbps (M.Knapex et alt 2006). In 2008 a 2.5 Gbps experiments using a 1W laser at 1064nmwith a BPSK modulation format between a LEO satellite and a ground station hasbeen demonstrated (E. Leitgeb et alt, 2009).FSO has been also used to establish links between satellites and aircrafts orbetween aircrafts or even between satellite, aircrafts and ground stations realizingan ad hoc optical broadband wireless airborne network. The LOLA (LiaisonOptique Laser Aroporte) programme, for example, in 2006 demonstrated the firsttwo way FSO link between the ARTEMIS GEO satellite and a Myster 20 airplaneflying at 9000m. At the receiver side an accurate hemispherical broadband pointingsystem and a CMOS sensor for detection and tracking, with a pointing accurancybetter than 1 micron rad, was used (Cazaubiel et alt, 2005). Real time data

communications, video and audio, demonstrated via a 50 Mbps transmission witha link acquisition time under the second in 2006 and 2007. Deep Space Communications: ultra-long distance can be reached with the FSO

system, thanks to the recent developments in this field, in order to allow the linkwith deep space. A great number of studies investigate about the beam divergenceand the geometrical loss to obtain the features to established high data rate FSOlink between Earth-Satellite, Earth-Moon, Earth-Mars and Earth to celestial bodieswithin the solar system. To obtain a detector able to work with very low power,new technologies propose devices such as low noise photon-counting detector tob eplaced on the planets in the form of fields array. The link budget description isbased on EIRP (effective isotropic radiated power), Space losses and PDE

(photodetection efficiency. Furthermore problems can arise from laser to opticscoupling and turbulence in high atmosphere if passed. Modulation formats arbased on pulse pattern modulation (PPM), at wavelengths ranging from 1064nmand 1550nm via YDFA technology (D. Caplan et alt, 2007). Dimensioning theseparameters, an FSO system can be used for deep space mission since the opticalbeam can cover large distances and go through the space to reach the destination(Harris Alan et al., 2006). In particular the MLCD project expected performancesare 1Mbit/s farthest Mars and 30 Mbit/s nearest Mars. Increased performances cansupport data rate up to 1Gbit/s maximum Mars distance, 100 Mbit/s Jupiter and10 Mbit/s Uranus (D. Boroson, C. Chen, B. Edwards, 2005).

www.intechopen.com


31/41


Undersea communications: FSO is used to transport information from fixed ormobile sea-platforms to other stations without deploying any cable. Using aparticular wavelength it is possible to cover small distances (around 100 m) butwith transmission rate around 10Mb/s. The disadvantage is that the sea offers an

attenuation greater than the attenuations offered by the air. Submarine LaserCommunications (SLC) is implemented to achieve high data rate transmissions alsoin case of emergency between two platforms or between a platform and an airbornesystem. SLC is used for communication with deeply submerged submarine usingan FSO laser at particular wavelength. Often it is used green or blue wavelengths,thus the radiation is placed in visible spectrum. For undersea application,particular type of lasers are used such as xenon chloride (XeCl) laser shifted in thevisible spectrum. Each submarine receiver, has the ad-hoc detector to capture theoptical laser beam. Several tests was made to study the performances of undersealaser link and also aircraft to submarine transmission system was implemented. Anaircraft flying at 40.000 feet was connected with a submarine using a doublewavelength: blue wavelength for uplink and green wavelength for the downlinkstream. There was also clouds between aircraft and submarine but the detectorinstalled on board of the submarine was able to detect the signal since a specialoptical receiver was applied. In this way, the use of the blue-green opticalwavelength for undersea applications confirmed the use of these opticalfrequencies in the sea field. The disadvantage of this kind of laser is that its timeoperational life is not so long; but the technology, today permit to have solid-statelasers. This kind of solution permit to have a longer operational life laser and alsowith its efficiency is improved. The smaller cost respect to normal gas lasers,permit also the use of these devices in deep space scenario.

Air to Earth communications: to prolong the band of the RF technology used todayfor several airplanes carrier to monitor, perform surveillance actions and for GIS(Geographic Information System) applications (Juan C. Juarez et al., 2006). NASAJPL, on this side, demonstrated a 2.5 Gbps FSO link between an UAV and a groundstation studying in particular the atmospheric fades and the problems related to thepointing systems (G. Ortiz et alt, 2003).

Inter island communications: With the DOLCE study, ESA funded project, aninter island free space communication has been demonstrated covering 142 kmbetween La Palma and Tenerife at 10 Mbps with a 1W MOPA (Master OscillatorPower Amplifier) using a 32 PPM (Pulse Position Modulation) and a simple Si-APD as receiver placed on a OGS (Optical Ground Station) (G. Baister et alt, 2009).

On the same link, the ROSA project performed experiments to investigate anoptical telemetry system for the mars sample return mission (T. Dreischer, 2008).

Inter optical communications subsystems: board to board and chip to chip opticalwireless interconnections become a reality in the last years (Hirabayashi K. et alt,1997). The main reasons have been the need to compensate the board to boardbottlenecks and to increase the backplane interconnections speed. Especially in chipto chip wireless interconnections, the main problem arise from design and packageissues. Other challenges are focused to the development of ultra low drivingdevices for VCSEL arrays, commonly used in board to board interconnections, andto the increase of alignment tolerance. Experiments demonstrated that is possible to

www.intechopen.com


32/41


establish more than 1000 channels per printed circuit board using a 1mm pitchoptical beam array having at 1Gbps per channel, thus a throughput per board up to1 Tbps (Hirabayashi, Yamamoto, Hino 2004).

6. Techno economic analysis (Tosi Beleffi Forin)

A correct approach, in the techno economic analysis of a FSO based scenario, must focus onthe main market drivers that, today, are mainly related to the civil telecommunications.Undersea, inter planetary, inter satellites or military based communications, infact, still arein the research or prototyping field where, everyone knows, the costs are not in principle,especially at the beginning, strictly taken into account for a mass production. This preambleis important to justify the fact that this sub chapter will be mainly devoted to the civiltelecommunication applications where the competition of different actors is todayincreasing the market portfolio thus lowering the overall costs.The demand for broadband infrastructures, mainly driven today by the request for newmultimedia applications, is pushing the Operators to implement specific strategiescharacterized by a continuous and slow migration to the so called FTTx family (Fiber to TheCurb/Cabinet/Building/Home) where the final step is constituted by the FTTH (Fiber ToThe Home). In the future, infact, is expected that we will have an exact replica of the PSTNnetwork but with fibres instead of the copper. Each end user will have a single or a pair offibres directly connected with the CO (Central Office).Today the main effort is, for what has been previously mentioned, devoted to thedevelopment of new burying strategies to lower the CAPEX that, in the fibre optical basedinfrastructures, are mainly due to the fibre installation. The installation costs of a 36 fibreoptical cable, in a typical urban area environment, are, for example, divided between the dig

(12%), the cable and the cable lying (14%) and the civil works for surface footway (74%)(A.L.Harmer, 1999). For these reasons a tremendous proliferation of new techniques hasbeen experienced: trench, micro trench, dig, micro dig, one day dig, Teraspan, aerial cables.Several Operators and Municipalities are today performing demo trials to demonstrate thepossibility to put the fibre cables in the sewer pipes, inside the urban lighting systems oreven in the gas pipes. All these different approaches can, in principle, reduce the installationcosts, depending of course on the particular case and/or situation, in average of a 30-40%respect to a standard trench approach.But what can happen if must be crossed a river, a railway or connected a neighbour island?In this scenario, which can be the role of the FSO?It should be considered that nothing is so simple as reported with the pen on the paper. In

the case of digs, still many problems are present especially under the regulatory point ofview and, most important, for the huge amount of authorizations that has to be requested tothe Regions, Provinces, Municipalities and Districts. Is not so simple, costless and fast,infact, taking a excavator and start to dig along a street. On the other side, it must be pointedout that gas pipes, urban lighting systems, sewer pipes, water systems can be in principleused to host optical fibres but still remain critical infrastructures, under the security point ofview, and so the fully access to them is still difficult.We start to understand that the development and the diffusion of the broadband to the enduser is not only a matter of digging the fibre. Is a more complex problem where mixedwireless and wired infrastructures can and must coexist. This to limit the digital divide,

www.intechopen.com


33/41


increase the broadband to all, taking care of the costs, both CAPEX and OPEX (OperationalExpenditures). Depending on the geography, infact, a particular technology can be betterthan a different one, being wired or wireless. This open the way to the implementation ofmixed infrastructures and to the deployment of different technologies like: WiMax, WiFi,

SDH Radio, LMDS, WDCMA/UMTS, UTRAN, GSM, FSO, Satellite, xDLS, Fibre, Coax.Considering the FSO technology as a point to point based system, we can start to define anddifferentiate it respect to other possible competitors like the Fibre, the Microwave links, theXDSL and the COAx (see table 5).

FSO Microwaveradio

OpticalFibre

Coaxialcable

xDSL

Speed Gpbs Mbps Independent Mbps Mbps

Installation Moderate Difficult Difficult Moderate Difficult

Uses P2P/P2MPshort and

long reach

P2P shortreach

P2P/P2MPshort and

long reach

Campus,multi

dropshortreach

Phone anddata, access

telecomsector

Advantages Price vsperformances,security

Speed vsinstallation

HugeAvailablebandwidth,security

Betterthanothercoppermedia.

Low cost, isalreadypresent.

Disadvantages Dependent onthe climaticconditions

Can beintercepted

Installationcosts

Costs Speedlimited byinterferenceand cable

qualitySecurity Good Poor Very good Good Good

Maintenance Low Low Low Moderate High

Skills Moderate High High Moderate Moderate

Table 5. Comparison between different P2P networks

From table 5 we can start to figure out, respect to other P2P technologies, the sector ofinfluence that can be covered by the FSO technology. In order to understand whichbroadband technology can be the most efficient in terms of CAPEX and OPEX, we have togo deeper in the problem considering also the following economic factors like: cost per line,average return per user, mean time before failure (MTBF), mean time to repair (MTTR),

warranty by vendor, upgradable characteristics, operation and maintenance costs,manufactured respect to which standard.Going deeper and deeper in the analysis we can make a simple calculation pointing out themain media/devices needed to set up a point to point link (see Table 6). In this case we cansee that the main characteristics of a point to point optical wireless link are to be wideband,easy and fast to install as well as low cost respect to the other technologies. The mix made bycost per bandwidth per easy to use/install is the winner. The main drawback is of coursedue to the climatic conditions encountered that limit, in case, the maximum distance/bit rateachievable. Adding a RF (Radio Frequency) link, increases the costs but increases also theavailability.

www.intechopen.com


34/41


TECHNOLOGY MAIN COSTS LICENCES NOTES

FSO From 13k @ 155 Mbps to19k @ 1.2 Gbps (up to 3.6km). From 22k @ 155 Mbps

(up to 5.7km) to 33k @ 1.2Gbps (up to 5.3km) (source:vendor).

Is not needed alicence. An unatantum per year is

due to the reference/ control PA.

Fast installation bothindoor and outdoor.Radio backup is needed

to increase theavailability up to99.999%.

RF From 20k to 30k dependingon the length (1-10km)working @ 18, 26 or 38 GHzcan transmit up to 300 Mbps.(source: vendor).

Is needed a licence.Example: For28MHz in the 7GHzbandwidth along20km costs around5k. (source: PA)

Fast installation, limitedbandwdith. High timewindow if is consideredthe time needed to havethe licence.

FIBRE In the case of a P2MP systemthat is the most cost effectivein terms of CAPEXs respect to

a pure FTTH. A standardGPON OLT (with 4 G-Ethernet ports), a 1:16 splitter,an ONU serving 48 VDSL2end users, 15 ONTs with 2GEthernet ports and VOIP[Everything compliant withthe ITU G.984 standard] has atotal cost of around 26k.(Source: vendor).Around 70/m for the digand 10/m for the cable in

urban environment. In ri

intech-free space optical technologies

Documents